comparison between in-situ surface measurements …...comparison between in-situ surface...

Comparison between in-situ surface measurements and global climate model outputs of particle light scattering

coefficient as a function of relative humidity María A. BURGOS1,*, Elisabeth ANDREWS2, Gloria TITOS3,4, Virginie BUCHARD5,6, Cynthia

RANDLES5,7, Alf KIRKEVAG8 , and Paul ZIEGER1

1Stockholm University & Bolin Centre for Climate Studies, Stockholm, Sweden 2University of Colorado, Boulder, USA

3Institute of Environmental Assessment and Water Research, Barcelona, Spain 4Andalusian Institute for Earth System Research, University of Granada, Granada, Spain

5NASA/Goddard Space Flight Center, USA 6USRA/GESTAR, USA

7ExxonMobil Research and Engineering Company 8Norwegian Meteorological Institute, Norway

Funded by US Department of Energy

*[email protected]

17th AeroCom meeting – 16th October, 2018

𝝈𝝈𝒔𝒔𝒔𝒔(𝑹𝑹𝑹𝑹,𝝀𝝀), strongly depends on RH

Aerosols and Climate

Aerosol Particle

Relative Humidity

HYGROSCOPICITY:

o Direct and indirect effects on the Earth’s energy balance o Scattering (σsp) and absorption of solar radiation and the

number of cloud condensation nuclei will be affected by aerosol concentration, size and chemical composition

Since aerosol particles can take up water, they can change in size and chemical composition depending on the ambient relative humidity (RH)

The effect of water uptake is relevant for climate forcing calculations as well as for the comparison or validation of remote sensing with in-situ measurements and for the improvement of Global Climate Models

SCATTERING ENHANCEMENT FACTOR

𝑓𝑓 𝑅𝑅𝑅𝑅, 𝜆𝜆 = 𝜎𝜎𝑠𝑠𝑠𝑠(𝑅𝑅𝑅𝑅, 𝜆𝜆)

𝜎𝜎𝑠𝑠𝑠𝑠(𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅, 𝜆𝜆)

Introduction Motivation Measurements Models Comparison Conclusions

Hygroscopicity in GCM’s

Figures from Mian Chin (NASA Goddard)

Fraction of aerosol optical depth (AOD) due to water in different models:

ECHAM5: global annual average 76% GOCART: global annual average 40%


Hygroscopicity in GCM’s

OPAC: Optical Properties of Aerosol and Clouds (Hess et al., 1998)

OPAC model generally higher than measurements especially for low-medium RH

Reason: OPAC growth factors for sea salt and sulfate components are too high. Revised growth factors for sea salt published in Zieger et al., 2017.

Figures from Zieger et al., 2013


Tandem Humidified Nephelometer PSI system:

NOAA system:


Humidifier Drier WetNeph

DryNeph

Aerosol

RH~20 – 95%

RH<40%

(Fierz-Schmidehauser et al., 2010)

Humidifier WetNeph DryNeph Aerosol

RH~20 – 95% RH<40%

Tandem Humidified Nephelometer

Zieger et al., 2011

• Humidograms can be parameterized with different equations:

Carrico et al., 2003: 𝑓𝑓 𝑅𝑅𝑅𝑅 = 𝛼𝛼 (1 − 𝑅𝑅𝑅𝑅)−𝛾𝛾

Zieger et al., 2010: Fit separately for RH>75% or RH<65% Titos et al., 2016: Several equations, some of them reproduce deliquescence

→ Problem for sea salt aerosols (deliquescence)


Hygroscopic particles grow or shrink monotonically with ∆RH

Deliquescent aerosols undergo sudden phase transition (hysteresis)

DoE funded project: “Evaluation and improvement of the parameterization of aerosol

hygroscopicity in global climate models using in-situ surface measurements” (2016-2019)

compare with GCM’s

HARMONIZED DATA SET

Red: PSI and SU measurements

- DoE/ARM sites, PSI sites and more - Covering 18 years

1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 Year


f(RH=85%):


Arctic > Desert Rural > Marine >

MERRA Aerosol Reanalysis (MERRAero):


• Buchard et al. (2015): “Using the OMI aerosol index and absorption aerosol optical depth to evaluate the NASA MERRA Aerosol Reanalysis”

• MERRA Aerosol Reanalysis: reanalysis for the satellite era based on a version of the GEOS-

5 model, radiatively coupled to the Goddard Chemistry, Aerosol, Radiation, and Transport (GOCART) aerosol module (bulk (mass) scheme).

• GEOS-5 -> run in replay mode using 6-hourly atmospheric analysis from MERRA • Aerosol species: dust, sea-salt, sulfates, organic and black carbon • Assimilation of bias corrected MODIS AOD observations at 550 nm every each 3 hours • Provides a aerosol gridded data set covering from 2002 to 2015

CAM5.3-Oslo • Kirkevåg et al. (2018): “A production-tagged aerosol module for earth system models,

OsloAero5.3 – extensions and updates for CAM5.3-Oslo ”

• Aerosol module: OsloAero5.3 implemented in the atmospheric component CAM5.3-Oslo of the Norwegian Earth System model (NorESM1.2)

Improvements: treatment of emissions, aerosol chemistry, particle lifecycle and aerosol-cloud interactions New features: improved aerosol sources, aerosol particle nucleation, secondary organic aerosol production, emissions schemes for sea-salt, DMS and marine primary organics…

• Model data availability → Daily values → Period: January – December, 2010


• Time coverage of model data and measurements are not coincident. For consistency, short-term campaign sites with only a few months of measurements are compared to the same months of the model data.

• Uncertainty in measurements between 20-30%, which has to be taken into account in the measurement-model comparison

Relative Frequency of Occurrence of f(RH=85%)

ARCTIC SITES


Measured CAM5.3-Oslo MERRAero

• Measurements show higher variability while models present a narrower distribution

• Measurements variability may be affected by the change of particle concentration along the year: Arctic haze in spring/new particle formation in summer/low concentration in winter (Tunved et al., 2013)

MARINE SITES:



MERRAero: f(RH) sistematically peaks at the same value, independent of the site characteristics



CAM5.3-Oslo does better in reproducing the observed shape, though it tends to overestimate the measured values



Urban-Mountain-Desert Sites


• Urban Sites: Models reproduce observed f(RH) for Granada and Nainital, but overestimate in Shouxian and Manacapuro




• Mountain site (Jungfraujoch): Model surface is not the same as measurement surface, so wouldn’t expect models to do well necessarily




• Mountain site (Jungfraujoch): Model surface is not the same as measurement surface, so wouldn’t expect models to do well necessarily

• Desert site (Niamey): models reproduce the measurements of f(RH) quite well


Rural Sites:

• Each model exhibits consistent peak values of f(RH=85%):

• ~2 for MERRAero • ~2.5 for CAM5.3-Oslo

• Models systematically

overestimate f(RH=85%) except for Melpitz (MEL), where the measurements peak is shifted towards larger values relative to the other sites

Median Values and 25 and 75 Percentiles

Arctic Marine Rural Urban Desert


MERRAero


CAM5.3-Oslo



• Underestimates f(RH=85%) observations for Arctic sites


MERRAero




• Exhibits similar values for most Marine and Rural sites


MERRAero





• Inconsistent for Urban sites


MERRAero





• Inconsistent for Urban sites

• Does well for Desert site


MERRAero


• Overestimates f(RH=85%) relative to

observations, but better reproduces the diversity of observations



CAM5.3-Oslo




• CAM5.3-Oslo overestimates f(RH=85%) for Arctic sites (opposite of MERRAero)



CAM5.3-Oslo





• Reproduces the diversity in Marine sites with a general overestimation



CAM5.3-Oslo






• Exhibits approximately constatn f(RH=85%) at Rural sites – does NOT capture observed diversity



CAM5.3-Oslo







• Inconsistent results for Urban sites, with a

tendency to overestimate



CAM5.3-Oslo







• Inconsistent results for Urban sites, with a

tendency to overestimate

• Does well for the Desert site



CAM5.3-Oslo

Annual Cycles


• GRW (Marine): • the value of f(RH=85%)=2 simulated by MERRAero is constant throughout the year • CAM5.3-Oslo simulates a similar cycle to the observations with a bias towards larger

values

• SGP (Rural): • both models overestimate f(RH =85%) throughout the year. • CAM5.3-Oslo tracks the observed annual cycle better than MERRAero

• BRW (Arctic): • Both models track observed annual cycle (higher in autumn, lower in spring)

Annual Cycles


• Differences suggests some seasonal chemistry that models are not reproducing

→ Possibility to compare model and measurement chemistry at some sites to further assess

→ Study how number, surface and volume size distributions affect scattering

Next Steps...

• Re-analysis of data from 26 sites measuring different aerosol types to built a benchmark, harmonized and reliable database

• Comparison of f(RH=85%) between measurements and model outputs (MERRAero and CAM5.3-Oslo) highlights that:

• Constraint values of the model output for several aerosol types • Overall, CAM5.3-Oslo reproduces better the variability of measurements while MERRAero

present less variability • The f(RH=85%) values are coincident with measurements for some sites • Differences in seasonal chemistry may not be well represented in models

• Optical closure studies can help to reduce uncertainties (not possible at all sites due to measurement restrictions)


• Study the covariance of aerosol hygroscopic growth with other intensive properties such as SAE or SSA

• Study what is considered a valid definition of “dry RH” and the changes in optical properties at low RH conditions and its implications (Poster Andrews, P02)

Questionaire to AeroCom modelling community to collect metadata and a description of growth parameterization Variables requested: • Aerosol extinction, 550 nm, 40%, 55%,

65%, 75%, 85% RH + ambient • Aerosol absorption, 550 nm, 40%, 55%,

65%, 75%, 85% RH + ambient • AOD speciated Years of simulation/emission: • 2010 • Optimal: 2000-2014

Please participate! Description of data request can be found at: https://wiki.met.no/_media/aerocom/INSITU_AeroComPIII_description.pdf

We encourage you to provide model data!!

REFERENCES: Buchard et al., 2015: Using the OMI aerosol index and absorption aerosol optical depth to evaluate the NASA MERRA Aerosol Reanalysis, Atmos Chem Phys Chin, M. et al., 2002: Tropospheric aerosol optical thickness from the GOCART model and comparisons with satellite and sun photometer measurements, J.A.S. Fierz-Schmidehauser et al., 2010: Measurements of relative humidity dependent light scattering of aerosols, Atmos meas tech Hess, M., et al., 1988: Optical Properties of Aerosols and Clouds: The Software Package OPAC, A.M.S. Kirkevåg et al., 2018: A production-tagged aerosol module for earth system models, OsloAero5.3 – extensions and updates for CAM5.3-Oslo, Geos Model Develop Randles, C. A. et al., 2013: Direct and semi-direct aerosol effects in the NASA GEOS-5 AGCM: aerosol-climate interactions due to prognostic versus prescribed aerosols, J.G.R. Tang, I. N. et al., 1997: Thermodynamic and optical properties ofvmixed-salt aerosols of atmospheric importance, J.G.R. Titos et al., 2016: Effect of hygroscopic growth on the aerosol light-scattering coefficient: A review of measurements, techniques and error sources, Atmos environ Zieger et al., 2010: Effect of relative humidity on aerosol light scattering in the Arctic, Atmos Chem Phys Zieger et al., 2017: Revising the hygroscopicity of inorganic sea salt particles, Nature Communications.

THANK YOU for your ATTENTION!

María Ángeles Burgos ([email protected])

Related poster: Andrews, P02

BackUp slides

María Ángeles Burgos ([email protected])


Implementation of hygroscopic growth (Randles, C. A. et al., 2013 ): • Carbonaceous species and sulfate: parameterized based on OPAC (Hess et al., 1998) as in

Chin et al. (2002) • Sea salt: parameterized based on observations of mixed-salt aerosol growth from Tang et

al., (1997)

MERRA Aerosol Reanalysis (MERRAero):

• Hygroscopic growth factors for aerosol components at some typical dry radii and for relative humidities up to RHmax = 99.5%

CAM5.3-Oslo

2009-2013

85%/40% N.data(2009)=[0,0,0,0,0,0,0,0,2,5,16,9] N.data(2010)=[19,43,58,15,10,0,0,20,19,19,16,49] N.data(2011)=[21,20,25,24,2,0,0,0,5,0,3,25] N.data(2012)=[19,34,41,0,2,4,2,3,5,4,12,40] N.data(2013)=[36,12,50,40,6,1,2,0,3,0,0,0]

Wet/dry N.data(2009)=[0,0,0,0,0,0,0,0,5,13,33,11] N.data(2010)=[51,78,39,17,0,1,36,28,32,19,52] N.data(2011)=[24,23,34,24,3,0,0,0,19,0,3,73] N.data(2012)=[34,63,63,0,11,11,5,4,15,26,19,60] N.data(2013)=[92,33,75,54,9,6,8,5,14,2,0,0]

Checking the time series of BRW for the measurements

MERRAero CAM5.3-Oslo



Median Values and Percentiles 25 and 75

Linear fit and 95% prediction interval (y±2∆)

SITE Measurements CAM5.3-Oslo

Appalachian 1.7±0.4 2.3±0.2

Barrow 2.4±0.6 2.5±0.3 Cabauw 2.2±0.6 2.5±0.3

Finokalia 2.5±0.6 2.3±0.5 Black Forest 1.5±0.4 2.3±0.3

Graciosa 2.3±0.6 3.0±0.3

Gosan 2.1±0.4 2.3±0.4 Shouxian 1.6±0.3 1.9±0.3

Hyytiälä 1.2±0.3 2.3±0.3

Jungfraujoch 2.3±0.8 2.3±0.3 Manacapuro 1.2±0.1 1.8±0.2

Mace Head 2.5±1.0 2.9±0.3

Melpitz 2.3±0.5 2.3±0.3

Niamey 1.3±0.5 1.3±0.1

Nainital 1.5±0.4 1.7±0.3 Cape Cod 1.9±0.5 2.5±0.2

Point Reyes 2.6±0.7 2.5±0.3 Southern Great Plains 1.7±0.6 2.3±0.2

Trinidad Head 2.0±0.7 2.6±0.4

Granada 1.8±0.4 2.4±0.4

Zeppelin 2.5±1.3 2.8±0.3


SITE Measurements MERRAero Appalachian 1.7±0.4 2.0±0.1

Barrow 2.4±0.6 2.0±0.1

Cabauw 2.2±0.6 1.9±0.1 Finokalia 2.5±0.6 1.9±0.2

Black Forest 1.5±0.4 1.9±0.1

Graciosa 2.3±0.6 2.1±0.1

Gosan 2.1±0.4 2.1±0.1 Shouxian 1.6±0.3 2.0±0.1

Hyytiälä 1.2±0.3 2.0±0.1

Jungfraujoch 2.3±0.8 1.8±0.1

Manacapuro 1.2±0.1 1.9±0.1

Mace Head 2.5±1.0 2.1±0.1

Melpitz 2.3±0.5 1.9±0.1

Niamey 1.3±0.5 1.2±0.1

Nainital 1.5±0.4 1.7±0.2

Cape Cod 1.9±0.5 2.1±0.1 Point Reyes 2.6±0.7 2.0±0.1

Southern Great Plains 1.7±0.6 2.0±0.1

Trinidad Head 2.0±0.7 2.1±0.1

Granada 1.8±0.4 1.7±0.2 Zeppelin 2.5±1.3 2.0±0.1


comparison between in-situ surface measurements …...comparison between in-situ surface...

Documents